Texas has spent the last two decades steadily cutting major air pollutants, even as its population and economy surged. That progress is now facing a new kind of stress test from a source that rarely appears in traditional environmental debates: the fast expanding data center industry. In a state that already leads the nation in power generation, the question of how electricity for these facilities is supplied could shape the future of Texas air quality and greenhouse gas emissions from data centers for years to come.

The Houston Advanced Research Center (HARC) recently examined how this surge in electricity demand might affect the state’s air. Their analysis focuses on the power generation and industrial sectors, which are directly tied to data center growth, whether those facilities draw power from the grid or rely on their own on-site generation. What emerges is a picture of hard-won air quality gains on one side and a potential turning point on the other, where choices about technology and fuels could either protect or undermine that progress.

From 2002 to 2024, Texas cut major pollutants like carbon monoxide, nitrogen oxides, and sulfur dioxide sharply, even as millions of new residents arrived and the economy roughly doubled in size. According to the assessment, that achievement was driven by a combination of cleaner power plants, stricter vehicle standards, and the retirement of coal-fired generation. Sectors most directly regulated at the federal level, particularly power generation and on-road transportation, made the most dramatic and sustained reductions by using enforceable technology standards at the point of manufacture or operation.

Those same sectors are now in the spotlight because they are intertwined with data center development. If a data center connects to the grid, its impact is tied to the evolving grid generation mix. If it installs on-site generation, such as natural gas engines or cogeneration systems, its emissions profile looks very different. The trajectory of power generation and industry, particularly under conditions of rapid growth, will play a critical role in determining Texas’s overall emissions footprint.

To understand these possible futures, HARC modeled four power supply scenarios across three projections of data center energy demand. The result was 77 distinct trajectories covering five pollutants over the next several years. These scenarios are presented in an interactive chart that allows policymakers and stakeholders to explore how different combinations of demand and power supply affect emissions.

The findings show just how powerful the choice of power supply can be. Holding data center demand constant, volatile organic compound emissions could be anywhere from 2 to 165 times higher if the new electricity needs are met with on-site combustion rather than grid supply paired with solar and wind. In the most optimistic cases, one year of emissions from data centers would be so small that they would be almost a rounding error compared with what Texas already emits. At the other extreme, for carbon dioxide and nitrogen oxides in particular, the additional emissions could exceed everything Texas currently emits in a year for those two pollutants combined. HARC’s assessment emphasizes that the difference between these outcomes comes entirely from power supply technology choices, which underscores the cumulative impact that a large wave of data centers could have.

The study is careful to note that these conclusions are not predictions. They are plausible futures based on current data and technology assumptions. Texas lawmakers, regulators, utilities, and industry leaders still have an opportunity to shape how this story plays out. The modeling makes clear that the scale of the projected data center build-out means decisions made now about electricity sourcing, on-site generation, and emissions controls could either keep air quality improvements on track or begin to reverse them.

Behind these scenarios lies a detailed methodology. Data center capacity is drawn from three sources: ERCOT’s Adjusted Large Load Breakdown, Transmission Service Providers’ reports, and industry data from BAXTEL. For BAXTEL, new, under construction, and planned capacity is added each year to the cumulative total. All data centers are assumed to operate continuously at 80 percent capacity. Their total annual electricity demand is calculated by multiplying that capacity by the number of hours in a year, then by 0.8. About 40 percent of the electricity is assumed to go toward cooling, in line with estimates from the Congressional Research Service, McKinsey, and Deloitte, and the cooling load is calculated using a coefficient of performance of 4.69.

For scenarios other than pure grid supply or on-site natural gas engines, HARC relied on the National Renewable Energy Laboratory’s REopt tool to simulate the electricity and fuel mix for a representative 1 gigawatt data center. This includes, for a net zero grid plus solar and wind scenario, how much power comes from each source. For reciprocating engines and cogeneration systems, federal factsheets and REopt estimates were used to size fuel usage. A carbon capture case was also modeled, where a post combustion system is tied to a cogeneration plant serving the same 1 gigawatt facility. Once the mix for this reference facility is set, fuel and electricity use for all projected Texas data center capacity is estimated by linear scaling.

Emissions factors for grid electricity combine data from EPA’s eGRID with ERCOT’s own generation data by fuel type. Capacity factors are calculated for each fuel using ERCOT’s historical production and capacity data, then projected forward using trends with a damping factor to avoid unrealistic extrapolation. Future generation by fuel type is then estimated and combined with eGRID plant level emissions factors to produce a future grid emissions factor. For natural gas used in on-site combustion, EPA carbon dioxide emissions factors based on higher heating value are applied as a constant across all years. Factors for nitrogen oxides, volatile organic compounds, fine particulate matter, and sulfur dioxide are taken from Texas and federal documents for natural gas reciprocating engines and gas turbines. The analysis accounts for both uncontrolled emission rates and scenarios with post combustion control systems that substantially reduce these pollutants.

To understand the additional effect of data centers, HARC first established a baseline in which no new facilities are added and Texas grid emissions continue along their recent trend. For pollutants such as nitrogen oxides, volatile organic compounds, sulfur dioxide, and fine particulate matter, the Texas Commission on Environmental Quality’s point source emissions inventory from 2018 to 2024 was used, and the share associated with the grid was estimated using EPA’s National Emissions Inventory. Average percentage changes in emissions over recent years are used to project trends beyond 2025. For carbon dioxide, a similar approach draws on federal data for Texas grid emissions. Together, these values form a baseline against which added data center emissions are compared.

Each scenario then layers data center emissions on top of this baseline to produce total projected emissions by pollutant from 2018 through 2031. For the year 2031, the minimum and maximum projected values and their percentage change from 2025 are summarized across all scenarios. The result is not a single forecast but a range of futures that show how sensitive Texas’s air quality trajectory is to decisions about how new electricity demand is supplied.

The overarching conclusion from the assessment is straightforward. Texas’s air quality gains since 2002 are real and were achieved largely through cleaner power generation. The rapid expansion of data centers will tie the future of that progress to choices about grid mix, on-site generation, and emissions controls. In lower emission scenarios, data centers are a manageable addition to the state’s inventory. In higher emission scenarios, they represent a potential turning point that could erase much of the improvement achieved over the past two decades. The window to influence which of those futures becomes reality is open now, and the modeling provided by HARC offers a data driven foundation for those decisions. Read more at https://harcresearch.org/research/texas-data-center-air-quality/